Building control with AI and sensor technology for efficiency

Technical building equipment has changed significantly in recent years. While individual systems used to be considered in isolation, holistic control is now increasingly taking centre stage.

Digital systems utilise existing infrastructure and supplement it with additional sensor technology. Such an approach is based on a combination of data collection, analysis and automated control. The aim is not only to monitor the operation of buildings, but to actively optimise it.

Baind combines Edge Box with existing infrastructure

The system is accessed via an edge box, which is installed directly in the building. Baind uses this component to read existing data from the existing infrastructure. It first analyses what network technology is available and what information is already available. The Edge Box accesses this data and integrates it into a higher-level system. This means that it is not necessary to completely replace the technology, but existing structures can continue to be used. This reduces the implementation effort and enables a quick start to data analysis. This first step forms the basis for all further functions, as it establishes the link between the physical system and digital analysis.

LoRaWAN sensor technology supplements missing data points in the room

In many buildings, the existing data is not sufficient to enable precise control. In such cases, additional sensors are installed. Baind relies on battery-operated LoRaWAN sensors, which can be retrofitted without major structural work. The sensors are usually installed directly in the individual rooms. They are often installed using simple adhesive solutions, which avoids interfering with the fabric of the building. With a battery life of five to ten years, the systems are designed for long-term operation. Various parameters that are relevant for the control system are recorded:

• temperature in the room
• Humidity and CO₂ values
• Lighting conditions
• Presence of people

AI control system processes data from the building and surroundings

The actual control takes place via an AI-based evaluation of the collected data. Not only internal measured values are taken into account, but also external influences. The additional data sources include, for example, weather information or data on in-house electricity production. The combination of these factors enables a differentiated view of building operation. The AI continuously analyses how the building is behaving and uses this to develop a model for efficient control. User behaviour is also taken into account. Occupancy patterns and typical usage times are included in the calculation. One practical example is the adjustment of operating times. If a high level of self-generated electricity is expected, the system can specifically start systems later in order to optimise the use of self-generated electricity.

Self-learning systems adapt to utilisation and season

Continuous adaptation is a key component of the system. The AI does not work with fixed specifications, but continuously develops its models. The first effects can be recognised after just a short time. The majority of potential savings are realised within the first two weeks. At the same time, accuracy improves with increasing runtime, as seasonal influences and recurring patterns are recognised. The system is regularly retrained. The models are updated at intervals of a few days and adapted to new conditions. As a result, the control system remains dynamic and reacts to changes in operation.

Energy savings through data-based optimisation

The combination of sensor technology and AI leads to measurable effects in energy consumption. In test buildings, savings in the range of 20 to 30 per cent were achieved, in some cases significantly more. These figures show the potential that lies in data-based control. The decisive factor here is not a single measure, but the interplay of many small adjustments during operation. The most important influencing factors are

• precise recording of actual usage behaviour
• Inclusion of external data such as weather or energy generation
• continuous adaptation of the control system using AI
• Integration of existing infrastructure without complete retrofitting

Business model and classification in the market

In addition to the technical implementation, the business model also plays a role. The solution is offered as a subscription-based service in which the AI is continuously developed and updated. The entry costs vary depending on the version, but are in the region of several thousand euros for high-performance systems. The offer is therefore primarily aimed at operators of larger buildings, where savings have a corresponding impact. With this approach, Baind demonstrates how existing buildings can be operated more efficiently using digital technologies. The combination of existing infrastructure, additional sensor technology and self-learning control is changing the way we look at building technology - away from static systems and towards dynamic, data-driven processes.